Hi I'm Aaron, I am majoring in Computer Engineering and Computer Science at NNU. I'm also the computer engineering lead for NNU's Rocksat-X program that collaborates with NASA, CU and other universities. I did research for NNU's Ag Computer Vision research project. This included researching different machine learning algorithms and techniques to detect and count blossoms in images. I am also the vice chair of NNU's IEEE chapter.

MARSHA is part of a research project at Northwest Nazarene University studying the use of Meta-Reinforcement Learning for robotics to account for inaccuracies between simulation and real-world, zero-gravity physics. The project includes developing a pair of robotic arms that will use Multi-Agent communication to maneuver objects in space by throwing and catching them. The use of meta-learning allows the robotic arms to complete the task after training in a simulated environment and after a few training attempts in the real-world zero-gravity environment. Therefore, once the robotic arms have been fully trained, they can be used in space to move objects large distances while using electric motors instead of propellant.

NNU's 2020 RockSat-X project's mission is to research radio frequency communication in space, disintegration of 3D printed plastics during re-entry, and virtual reality experiences in space using a 360 degree camera. My task involved designing a system using an Arduino and Raspberry Pi that can communicate via Bluetooth and send signal strength as well as sensor data from the end of a scissor boom. I also designed the master script using python's threading module that will activate and control several devices such as a GoPro, DC motor, Servo motors, 360 degree camera, and the RF experiment when signals are detected from the rocket. I also designed a system to send the signal strength and sensor data back to NASA through the rocket's telemetry lines.

An AI that predicts stock prices using an RNN trained on thousands of bars of stock market data. The AI then uses that prediction to buy stocks with the greatest reward and lowest risk. Currently upgrading the RNN to a reinforcement learning technique using a Deep Q RNN which is capable of buying and selling at the opportune time. Click here to visit the Kara website.

DragonFly is a quadcopter capable of shooting targets with foam balls. It is controlled with a Raspberry Pi. Its targeting system uses a CNN to detect targets. It utilizes a PID controller for stability. When complete the DragonFly will be able to autonomously search an area for targets and hit them with the foam balls. A safety system is being developed to ensure misfires do not occur.